Contemporary Issues in Risk Assessment

June 17, 2015

Identification and Selection of the Evidence Base for Human Health Assessments

Kate Z. Guyton, PhD DABT Senior Toxicologist Monographs Programme International Agency for Research on Cancer World Health Organisation Lyon, FRANCE

Conflict of Interest Statement

I declare no financial interests related to the subject matter of my presentation.

Outline

Background to systematic review A protocol for cancer hazard identification: the

Preamble to the IARC Monographs Identifying and selecting relevant literature:

experience using HAWCproject.org – Case example: volume 112 (March, 2015) – Capturing identified studies into tabular and narrative

summaries Future opportunities and conclusions

Key concepts: Hazard vs. Risk

Evaluate risks: the probability that cancer will occur, taking into account the level of exposure to the agent

Hazard Identification Risk assessment

Identify causes of human cancer: chemicals, complex mixtures, occupational exposures, physical and biological agents, lifestyle factors

Key Recommendations: Evidence Identification, Evaluation, Synthesis

• Document studies identified, excluded and included • Use templates for evidence display tables • Establish protocols for evaluation and synthesis

US National Research Council:

World Health Organization:

“Evidence, evidence, evidence”

IARC Monographs Preamble: Cancer Hazard Identification Protocol

http://monographs.iarc.fr/ENG/Preamble/index.php

A. GENERAL PRINCIPLES AND PROCEDURES 1. Background 2. Objective and scope 3. Selection of agents for review 4. Data for the Monographs 5. Meeting participants 6. Working procedures

B. SCIENTIFIC REVIEW AND EVALUATION 1. Exposure data 2. Studies of cancer in humans 3. Studies of cancer in experimental animals 4. Mechanistic and other relevant data 5. Summary 6. Evaluation and rationale

Scientific Review and Evaluation

Overall evaluation

Cancer in humans

Cancer in animals

Mechanisms

Systematic reviews of human, experimental, and mechanistic data are considered together in overall evaluations

How to Identify Relevant Published Studies?

• Literature collected by IARC; meeting participants are expected to supplement the IARC literature searches with their own searches

Considerations: 1. Monographs cite 100s to 1000s of studies 2. Evolution in experience over time:

• Mail box(es) of papers (1970s-1980s era) • Electronic reference list, PDFs, indexed reference

database, MyNCBI searches (early 2000s) Challenges: 1. How, when, where were searches performed?

• So many mechanisms, so little time: how to search systematically?

2. How to capture studies from “hand searching”? 3. Which studies were included/excluded, and why?

Outline

Background to systematic review A protocol for cancer hazard identification: the

Preamble to the IARC Monographs Identifying and selecting relevant literature:

experience using HAWCproject.org – Case example: volume 112 (March, 2015) – Capturing identified studies into tabular and narrative

summaries Future opportunities and conclusions

Identifying and Selecting the Literature

Overall evaluation

Cancer in humans

Cancer in animals

Mechanisms

How, When, Where Were Searches Performed: Cancer in Humans

How, When, Where Were Searches Performed: Cancer in Animals

Inclusion/Exclusion of Studies: Cancer in Humans, Cancer in Animals

Using “Tags” to track disposition of each identified study • Function as exclusion criteria for any excluded studies

• Document the evidence stream(s)

pertinent for included studies

• Can be applied by Working Group (v112) or Secretariat for further Working Group review (v113)

Visualization- Included/Excluded Cancer in Humans, Cancer in Animals

Cancer in Humans

Cancer in Animals

Identifying and Selecting the Literature

Overall evaluation

Cancer in humans

Cancer in animals

Mechanisms

Mechanistic Studies: Special Challenges

Insights from Volume 100 and Recent Advisory Groups: • Monographs consider representative studies

to give a concise description of the relevant data and issues

• Increasing volume and complexity of mechanistic literature

• Systematic identification of mechanistic data is needed (i.e., pertinent to 10 key characteristics of carcinogens)

• Analysis of high-throughput/-content data (including from curated government databases) is encouraged

How to Systematically Identify Mechanistic Studies?

Source: ME Kushman, A Kraft, KZ Guyton, WA Chiu, SL Makris, and Ivan Rusyn. A systematic approach for identifying, evaluating, and presenting mechanistic evidence in human health assessments. Regul Toxicol Pharmacol. 2013 Nov;67(2):266-77.

How, When, Where Were Searches Performed: Mechanisms

And similar searches, to cover 10 “key characteristics”

How to Capture Studies from “Hand Searching”?

Prior evaluation in Volume 30 (1983)

Visualization- Included/Excluded Mechanistic Studies

Included studies organized by topic, species per “Instructions to Authors”

What about Additional Sources of (Publicly Available) Data?

Cancer in Humans: Published studies: Public “call for data” may identify recent publications Cancer in Animals: “Data from governmental reports that are publicly available” (US NTP,

Japan JBRC, etc): Must provide sufficient detail for independent assessment

Mechanisms: Published studies (frequently voluminous) “Data from governmental reports that are publicly available”

o Included in bioassay reports or databases (e.g., ToxRefDB) o High-throughput testing databases (e.g., Tox21)

Outline

Background to systematic review A protocol for cancer hazard identification: the

Preamble to the IARC Monographs Identifying and selecting relevant literature:

experience using HAWCproject.org – Case example: volume 112 (March, 2015) – Capturing identified studies into tabular and

narrative summaries Future opportunities and conclusions

Step 1: Making an Outline

• Online publication tools can facilitate contributions and peer reviews from multiple authors

• Assignments reflect topics, amount of literature to be covered, expertise

• Many other options: Open Monograph (open source) SharePoint Structured folders on (shared) drive/cloud HAWCProject.org- direct link to dose-response

Step 2: Including All Relevant Studies into the Database

Bulk upload of HAWC “included” studies (with links to PubMed)

Manually add references to “government reports”

Step 3: Developing Tabular and Narrative Summaries

Overall evaluation

Cancer in humans

Cancer in animals

Mechanisms

Capturing Data into Tables: Cancer in Humans

Pre-defined, required/optional fields with drop-down lists

Working Group comments captured

Ability to toggle between numbers and plots

Capturing Data into Tables: Cancer in Animals

Capturing Data into Tables: Mechanistic Data

Results can be sorted by any field

Dropdown options change according to assay system, endpoint

Dose, duration can be captured

Narrative Summaries: Address Pre-Specified Decision Criteria

Cancer in experimental animals

— Preamble Part B, Section 6(b)

Causal relationship has been established through either: - Multiple positive results (2 species, studies, sexes of GLP) - Single unusual result (incidence, site/type, age, multi-site)

Data suggest a carcinogenic effect but: (e.g.) single study, benign tumours only, promoting activity only

Studies permit no conclusion about a carcinogenic effect

Adequate studies in at least two species show that the agent is not carcinogenic

Conclusion is limited to the species, tumour sites, age at exposure, and conditions and levels of exposure studied

Cancer in humans

Mechanistic and other relevant data

� Evidence suggesting lack of carcinogenicity

� Sufficient evidence

� Limited evidence

� Inadequate evidence

The Final Step: Integrating Evidence to Reach Overall Conclusion

Sufficient Limited Inadequate

EVIDENCE IN EXPERIMENTAL ANIMALS

Group 1 (carcinogenic to humans)

EVID

ENC

E IN

HU

MA

NS

Group 2A (probably carcinogenic)

Group 3 (not classifiable)

Group 2B (possibly carcinogenic) (exceptionally, Group 2A)

Group 2B (possibly carcinogenic)

Sufficient

Limited

Inadequate

32

Group 1

Group 3

Group 3 4 consistently and

strongly supported by a broad range of mechanistic and other relevant data

Group 4

2A belongs to a mechanistic class 2B with supporting

evidence from mechanistic and other relevant data

Group 3

2A belongs to a mechanistic class 2B with strong

evidence from mechanistic and other relevant data

Group 3

The Final, Final Step: Bringing in Mechanistic Data

Sufficient Limited Inadequate ESLC EVIDENCE IN EXPERIMENTAL ANIMALS

2A belongs to a mechanistic class where other members are classified in Groups 1 or 2A

Group 2B (exceptionally, Group 2A)

ESLC

Limited

Sufficient

Inadequate

1 strong evidence in exposed humans

Group 2A

1 strong evidence in exposed humans

2A strong evidence … mechanism also operates in humans

Group 2B 3 strong evidence

… mechanism does not operate in humans

EVIDENCE IN HUMANS

Summary Systematic review is fundamental in cancer hazard

identification The Preamble to the IARC Monographs is a

published protocol for who, what, how and when (and even where!) evaluations are conducted

On-line tools can aid: – Identifying and managing a voluminous and complex

scientific literature – Alignment of tabular presentations with “strength of

evidence” conclusions

Acknowledgments

34

SOT/FDA Organizers Betty Eidemiller (SOT) Suzy Fitzpatrick (FDA) Allen Rudman (FDA)

Andy Shapiro, MS (NIEHS/NTP)

The IARC Monographs Volume 112 Working Group

The IARC Monographs Staff